Violence toward health care workers is a significant occupational health concern. In one study of psychiatric staff, 73 percent reported having been assaulted at least once during their career and 28 percent reported being assaulted at least four times. Aside from the cost to the institution, injuries to staff cause significant emotional, social, biophysical and cognitive responses in the victim. And yet, despite the large body of research related to institutional violence in general, there is a notable lack of theoretical work needed to provide a scientific base for nursing interventions with potentially violent psychiatric patients. The purpose of this study is to investigate, using grounded theory methodology, the day-to-day strategies that nursing staff use to prevent psychiatric patients from escalating to violence. The specific aims are (1) to identify and describe nursing interventions that are used to de-escalate psychiatric patients who are escalating out of control. (2) to describe the context and conditions under which specific interventions are used to assist patients to regain control and (3) to construct a mid-range theory of the de-escalation process. The interactions between the nursing staff and the patients will be observed and both nursing staff and patients will be formally and informally interviewed. The data will be analyzed and a theoretical model will be constructed using the constant comparative method of grounded theory. The long-term goal is to use this theory of de-escalation as the basis for developing and testing innovative nursing interventions for preventing violence and/or restraint use in psychiatric inpatient units and other settings that treat potentially aggressive patients. This outcome would not only make a unique contribution to psychiatric nursing practice but would effectively decrease the costs to the institution and the use of potentially harmful interventions with patients.